On Wed, May 17, 2006 at 11:18:45PM +0200, Matijs van Zuijlen wrote: > On Mon, May 15, 2006 at 03:35:31AM +0200, Michael Biebl wrote: > > So, wpa_supplicant is the way to go. That is why I won't change NM to > > make wpa_supplicant support optional because I don't want to divert that > > much from upstream. So I tagged this bug wontfix. > > Agreed. The problem for my case simply is that wpa_supplicant supports less > drivers than libiw. I'm sure this will change.
I have now 'fixed' wpa_supplicant's wext-driver to support the airport card. Part my fix it is a bona fide bug fix, since the WEP key setting method was broken or at least sufficiantly different from iwlib's 'reference implementation' that it didn't work. I will try to make that one into a patch. The other part is a hackish attempt at stopping the wpa_supplicant driver from issuing ioctls related to WPA when the kernel driver doesn't support them. That needs some work. In any case, my card now works when I use wpa_supplicant on its own with a configuration file. In combination with nm, it still doesn't work, but debugging is a little difficult since I seem to be missing part of wpa_supplicant's output in nm's log. -- Matijs van Zuijlen http://www.matijs.net/
signature.asc
Description: Digital signature